The question of the relationship of Common Celtic to the other Indo-European languages remains open. For some time, it was held that Celtic stood in an especially close relation to the Italic branch; some scholars even spoke of a period when an Italo-Celtic “nation” existed, toward the end of the 2nd millennium bc. The existence of a q–p relationship (see above) inside Italic too (e.g., Latin quattuor “four,” but Oscan petora) was thought by some to support this view. Much of this argument is, however, based on accidental resemblances (e.g., the Irish future tense in f- and the Latin future in b-) or on formations such as the deponent and passive verb forms ending in -r, which at one time were known mainly in Italic and Celtic but have since been found in the Hittite and Tocharian languages as well. The undeniable common features between Celtic and Italic, such as the superlative endings of adjectives (Latin -issimus; Celtic *-samos, *-isamos), are hardly sufficient to justify the assumption of a special relationship, and the whole concept of an Italo-Celtic unity has been powerfully criticized by the linguists Carl J.S. Marstrander and Calvert Watkins.

The original home of the Celts cannot be located precisely, but, on the whole, the evidence points to the eastern part of central Europe. There is more evidence for their contacts with other Indo-European peoples. One group of Celts, at least, found themselves neighbours of the Germanic peoples and were often confused with them by classical writers. It can be inferred that the Celts had attained a higher standard of social organization than the Germanic peoples from the existence of words such as Gothic reiki and andbahts (modern German Reich, Amt), apparently borrowed from Celtic *rīgion “kingdom” and *ambactos “officer.” To the Greeks and Romans, on the other hand, the Celts were inferior in culture; Celtic words in Greek are restricted to those describing Celtic institutions, such as bardoi “poets.” The borrowings from Celtic into Latin, which derive mainly from the period before the expansion of Roman power, belong to a few restricted categories, such as war (lancea “lance”), transport (carrus “baggage wagon” and carpentum “light carriage”), and agricultural products (cervesia “beer”). When the Romans finally conquered Gaul and imposed their language, a number of Celtic words came into Latin, but the Celtic terms were mainly concerned with rural life. These are more common in French dialects than in standard French, which preserves a mere handful, such as mouton “sheep,” ruche “beehive,” and arpent “land measure.”

Early records of Celtic

Continental Celtic

Celtic died out very quickly in eastern Europe. In its farthest outpost in Asia Minor, it may be assumed that the Letter of Paul to the Galatians was addressed to a people whose culture was already Greek but whose Celtic origins are clear from names preserved by classical authors; e.g., Drunemeton “the very sacred place,” formed from two distinctively Celtic elements. When commenting on that Letter, St. Jerome (died ad 419/420) said that the Galatians still spoke a language almost the same as that of the people of Trier. As he had been in both places, his evidence cannot be dismissed offhand, and it may be that a few speakers of Celtic still existed in both areas. Since St. Jerome did not claim to know any Celtic dialect, however, his statement cannot be accepted with certainty. Speaking generally, the history of Continental Celtic comes to an end as that of Insular Celtic begins.

Insular Celtic

The earliest evidence for Insular Celtic consists, like that for Continental Celtic, mainly of names recorded by Greek and Latin authors. In the case of Ireland, these were entirely by hearsay, and many of the Irish place-names recorded by Ptolemy in the 2nd century ad have not yet been identified. From perhaps the 4th century, ogham inscriptions (see alphabet) are found in Ireland, consisting almost entirely of personal names. From the 5th century onward, British names in Latin inscriptions are recorded in Wales, as well as Irish names in both Latin and ogham alphabets in areas of Irish settlement. These scanty records are of value above all in establishing that, up to very nearly the time at which written documents become available, British and Irish had remained similar in structure to Gaulish. Thus, in Britain is found the genitive (possessive form) Catotigirni, which in Old Welsh gives Cattegirn, and in Ireland the genitive Dovatuci exists, which in Old Irish gives Dubthaich. These changes—loss of final syllables and connecting vowels, weakening of consonants between vowels, and so on—are very similar to what was happening to Latin in France at the same time; e.g., Latin avicellus and aqua finally became oiseau and eau. There is no satisfactory explanation of why these profound changes should have occurred at this time, nor can the period at which they occurred be fixed precisely, for the engravers of the inscriptions clearly went on using traditional forms long after the sound had changed.

Linguistic characteristics of the Insular Celtic tongues

The new languages, the only forms of Celtic that are known thoroughly, present a considerable number of unusual features, some of them unknown to other Indo-European languages. Some scholars have argued that these features may have resulted from the presence of a large non-Celtic substratum in the British Isles. Because it is hardly likely that the Celtic invasions of those islands began much before 500 bc or that the invaders exterminated the existing inhabitants, such a possibility cannot be denied. On the other hand, some features once thought to be exotic, such as the initial position of the verb in the sentence, have been convincingly demonstrated to be organic developments from Indo-European. Others, such as the system of counting by 20s, are clearly innovations, but this system is shared by English (“three score and ten”), French (quatre-vingts “80”), and Danish, in all of which it is also an innovation, as well as Basque, in which it appears to be old.

Phonological characteristics

The most remarkable phonological feature of Insular Celtic is the development of a double series of consonants in which strongly articulated consonants are distinguished from their weak counterparts. The two series were originally merely phonetic variants, with the strong variety occurring in absolute initial position and in certain consonant clusters and the weak elsewhere. Later, however, the two series became independent, or phonological. In the languages as they first appear in writing, considerable changes have taken place in the phonetic forms of the two series. Both in Irish and in Welsh, Cornish, and Breton, the opposition (contrast) of strong:weak in the voiced stops has been replaced by stop:spirant (e.g., b:v). (A spirant, such as v, f, s, is produced with local friction and without complete stoppage of the breath stream.) Irish has the same system for the unvoiced stops (e.g., t:th), but Welsh, Cornish, and Breton have voicing in this instance (e.g., voiceless t:voiced d). These changes by themselves are not very different from the weakening of consonants between vowels that occurs in other western European languages (compare Welsh pader “prayer,” a loanword from Latin pater “father,” with Spanish padre “father,” deriving from Latin patrem), but, in Insular Celtic, they occurred not only inside the word but also inside the phrase, so that the initial consonant of a word preceded by another word ending in a vowel was weakened. When the final syllables were lost in the evolution to the modern languages, these variations remained, and a system of initial mutations (changes) was thus set up. If, for example, a Goidelic nominative form *sindos kattos koilos “the thin cat” is reconstructed, this will give Old Irish in catt coel after the loss of final syllables, but the genitive *sindī kattī koilī “of the thin cat” will give in chaitt choíl with changed initial consonants. The same sort of change occurred in one Italian dialect: in Tuscan, there occur porta “door,” la forta “the door,” tre porte “three doors,” from Latin porta, illa porta, tres portae. In both cases, consonant weakening has spread from word to sentence; there is a common development, but it cannot be claimed that it is distinctively Celtic.

Grammatical characteristics

Another feature of Insular Celtic is its lack of the infinitive form of the verb found in most other Indo-European languages—e.g., English “to do,” “to call.” The equivalent is the verbal noun, which is a noun closely linked to the verb, though not necessarily derived from the same stem. Being a noun, it can have a following noun in the genitive case, which, in the older languages at least, is subjective or objective according to whether the verb with which it is linked is intransitive or transitive. Thus, from the Old Irish sentence téit in ben “the woman goes,” the verbal noun phrase techt inna mná “the coming of the woman” can be derived, whereas from marbaid in mnaí “he kills the woman” can be formed marbad inna mná (lais) “the killing of the woman (by him).” Among many other functions of the verbal noun is its use, when preceded by the appropriate preposition, with the substantive verb to provide a tense with continuous meaning. Thus, to téit in ben there is a parallel a-tá in ben oc techt “the woman is at going” (= “the woman is going”), and to marbaid in mnaí corresponds a-tá oc marbad inna mná “he is killing the woman.” The close resemblance of this system to that of modern English, in which it is a comparatively recent development, has been variously explained as the working of a substratum or, more recently, in terms of areal (regional) development.

Modern languages of the family

The discussion of the individual languages that follows divides them into the two main groups, beginning with Irish, which is the oldest attested.